Dear John, your basic logic is valid but the way you "estimate" the cross section for the LSP production by the time reversal of the decay is way too emotional.
The reaction by which you expect charged dark matter to be created everywhere, namely
$${\rm SM stuff} + {\rm LSP}_{\rm from\,dark\,matter} \to {\rm charged\, superpartners}$$
is indeed possible and it is essentially what the underground detectors of dark matter are trying to find. However, this reaction is extremely rare because dark matter, while everywhere, is very diluted - less than half a GeV per cubic centimeter
http://arxiv.org/abs/0907.0018
With this - relatively to nuclear densities - tiny mass density (even smaller than the cosmological constant), it takes some time to detect the collisions with the dark matter.
You may also misapply the time reversal. While the laws of physics produce CPT-invariant - and approximately T-invariant - Lorentz-invariant amplitudes, the actual probabilities of the opposite processes contain different kinematic factors: that's implied by the time asymmetry of mathematical logic, an elementary fact that - as I found out - a huge number of people misunderstand.
In particular, the decay of the charged superpartner to the LSP and a charged standard model particle is described by the "decay rate" while the opposite process is described by a "cross section". The cross section and the decay rate don't even have the same dimension (units). So it's obvious that they can't be "equal". And saying that one of them is "small" if the other is "small" is only true if you properly "normalize" what you mean by the vague adjective "small". You apparently didn't do it right. The sentence is surely not right if "small" is meant as "small for purposes I find practical".
The "decay rate" and the "cross section" of the two processes related by the time reversal differ by kinematic factors comparable to powers of the total energy of a particle in the process. If the decay rate (width) is e.g. $1 {\rm GeV}$, which makes the particle decay almost immediately in the detector, and if its mass is $100 {\rm GeV}$, the cross section, the corresponding cross section might be comparable to $1 {\rm GeV}/ (100 {\rm GeV})^3$ which is a tiny cross section. It was made tiny by the huge denominator, and the denominator is huge because the mass of the superpartner is high.
The two processes differing by "time reversal" have different prior probabilities. There is simply no "simple" way to relate the opposite probabilities. In a similar way, if glass "certainly" breaks down if it falls to the floor, it doesn't mean that it's likely that the pieces will "often" collect themselves and create a perfect unbroken glass. The opposite process is suppressed, relatively to the first one, by $\exp(-\Delta S)$, the exponential of the entropy difference between the initial and final state.
But even for microscopic processes, the relevant "probabilities" of the process have - aside from the shared $|{\mathcal M}|^2$, the squared Lorentz-invariant amplitude, lots of kinematic factors - namely $1/2E$ from both initial and final particles (which is symmetric with respect to the time reversal); and the integration over $d^3 p$ of final particles (not initial ones, which is time-reversal-asymmetric). So expecting that the probabilities of the time-reversed evolutions are "equal" is a trivial logical fallacy. They're not equal - that's not what T-invariance or CPT-invariance mean.
